Author: Andrew Dados
Date: 09:36:57 03/28/00
Go up one level in this thread
On March 28, 2000 at 11:42:10, Keith Conary wrote: > >>Sigh. Why has everyone to invent all the weels over and over. There are a lot of >>this type of facts that could be easily more or less scientifically established >>and written down. I suspect a lot of these matters are program independant. >>Yet... >> >> >>Regards, >>Bas Hamstra. > > >Is there some sort of source among chess programmers that contains algorithms, >schemes and even code that has been already created and proven? To prevent this >very problem of recreating the wheel everytime someone encounters a problem? If >not, it would be very useful and beneficial to the chess community and it would >allow industry standards to evolve and improve chess programs on the whole in >the future. > > > Keith I am afraid many inventions are not 'universal' e.g. they may or may not work for another program. Example for me is 'history heuristic' which doesn't improve my move ordering on average (in some cases it does, in others it worsens comparing to my schema of move ordering). I read that Phalanx uses 'response killer moves' which I also tried and didn't work for me. Search process should also fit qsearch model (in a way that it lessens horizon effect somehow, so for example making checks in qsearch may allow you nullmove straight into qsearch, while not making checks in qsearch would force you to stop nullmoving at some depth...). So I really don't think you can talk 'industry standarts' here - what we are trying to do is more 'modeling' a better chess player, not 'uniforming'...:) -Andrew-
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.